In very large distributed databased (e.g., NoSQL databases) in which data is partitioned across multiple servers, retrieving the first N items with a given sort order can be very slow because the data is typically stored in the database unsorted. Therefore, the database must retrieve the items and then sort the items before returning them to the requestor. For large data sets, this is impractical or prohibitively slow. Improvements in the area are needed.